Project: Traffic Signals Safety and Reference: 232984

Efficiency Project Prepared for: RAC WA Revision: 2 Transport Analysis Report 9 October 2013

Contents

1. Introduction 5 1.1 Project purpose 5 1.2 Approach 5 2. Perth Context 6 2.1 Governance 6 2.2 Traffic signals system 6 3. Methodology 7 3.1 Stage One – Initial investigations 7 3.2 Stage Two – SCATS options testing 8 4. Site Overviews 9 4.1 Tonkin Highway/ Kelvin Road 9 4.2 Orrong Road 12 5. Model Build and Calibration: Existing Situation 20 5.1 Overview 20 5.2 Data sources 22 5.3 Assumptions 22 5.4 Model calibration and validation 23 6. Option Testing: Tonkin Highway/ Kelvin Road 27 6.1 Option plan 27 6.2 Option testing results 27 6.3 Summary results 29 7. Option Testing: Orrong Road 30 7.1 Option plan 30 7.2 Option testing results 30 7.3 Summary results 34 8. Key Findings and Conclusions 35 9. Recommendations 40

Appendices

Appendix A Further Crash Data

Project 232984 9 October 2013 Revision 2 Page 1

Appendix B Calibration Details Appendix C Modelling Delay Outputs Appendix D SCATS Counts Comparison

Index of Figures Figure 4-1: Tonkin Highway/ Kelvin Road site location map 9 Figure 4-2: Tonkin Highway/ Kelvin Road site location map 9 Figure 4-3: Tonkin Highway/ Kelvin Road intersection layout 10 Figure 4-4: Orrong Road looking northwest to Oats Street 13 Figure 4-5: Orrong Road site location map 13 Figure 4-6: Orrong Road/ Francisco Street intersection layout 14 Figure 4-7: Orrong Road/ Archer Street intersection layout 14 Figure 4-8: Orrong Road/ Wright Street intersection layout 15 Figure 4-9: Orrong Road/ Oats Street intersection layout 15 Figure 4-10: Orrong Road corridor (Francisco Street to Oats Street) 17 Figure 5-1: Model area for Tonkin Highway/Kelvin Road 20 Figure 5-2: Model area for the Orrong Road corridor 21 Figure 7-1: Eastbound Journey Times for the Orrong Road corridor in the AM peak hour period 32 Figure 7-2: Westbound Journey Times for the Orrong Road corridor in the AM peak hour period 32 Figure 7-3: Eastbound Journey Times for the Orrong Road corridor in the PM peak hour period 33 Figure 7-4: Westbound Journey Times for the Orrong Road corridor in the PM peak hour period 33 Figure 8-1: Main Roads WA signal adjustment process 36 Figure 8-2: Signal adjustment process used for this study 37

Index of Tables Table 4-1: Total number of crashes 2008-2012 11 Table 4-2: Crash locations at signalised intersections 17 Table 4-3: Total number of crashes from 2008-2012 18 Table 4-4: Crash ranking comparison for Orrong Road corridor 18 Table 5-1: NSW RMS Criteria for Turn Counts at Tonkin Highway/ Kelvin Road 24 Table 5-2: NZTA EEM Criteria at Tonkin Highway/ Kelvin Road 24 Table 5-3: NSW RMS Criteria for Turn Counts for the Orrong Road Corridor 25 Table 5-4: NZTA EEM Criteria for the Orrong Road Corridor 25 Table 5-5: Westbound Journey time comparison for the Orrong Road corridor 26 Table 5-6: Eastbound Journey time comparison for the Orrong Road corridor 26 Table 6-1: Tonkin Highway/ Kelvin Road total vehicle delay (hours) 28 Table 6-2: Tonkin Highway/ Kelvin Road total vehicle delay (percentage change from base) 28 Table 6-3: Tonkin Highway/Kelvin Road average speed (km/h) 28 Table 6-4: Tonkin Highway/Kelvin Road carbon emissions (grams) 29

Project 232984 9 October 2013 Revision 2 Page 2

Table 7-1: Option test plan 30 Table 7-2: Orrong Road Corridor total vehicle delay (hours) 30 Table 7-3: Orrong Road Corridor total vehicle delay (percentage change from base) 31 Table 7-4: Orrong Road average speeds (km/h) 31 Table 7-5: Orrong Road Corridor carbon emissions (grams) 31

Glossary

Call back - When a phase is run twice within a single cycle.

Cycle - One complete sequence of the signal phases

Double Diamond Overlap - A traffic signal phasing arrangement for a cross-roads intersection consisting of two right turn phases controlling the right turn movements off all approaches. Each right turn phase consists of both opposing right turn movements proceeding at the same time but also includes an overlap option. The overlap option occurs when one of the right turn movements has no more right turning traffic remaining. This right turn terminates through a yellow and red arrow display after which the opposing through movement is given a green. While this is occurring, the other right turn movement can continue to turn right. This right turn phase operation is repeated on the other two opposing approaches, hence the “Double”.

Exclusive Pedestrian Phase - When all vehicle lanterns are red and all pedestrian lanterns are green.

Flexilink - The local traffic signal controller uses predefined plans and time of day schedules for cycle time, phase splits, and offset from the fixed point. Each signal controller uses a software clock synchronised with the other intersections it coordinates with.

GEH - A formula derived by Geoffrey E Havers that is used to compare two independent sets of data while considering the size of values. The lower the GEH value, the closer the fit.

Incremental Split Selection - A routine used by SCATS to enable phase splits for up to four phases to be incrementally adjusted between +4% and -4% each cycle based on the degree of saturation of chosen stopline detectors. The alternative is that SCATS can only choose from predefined phase split plans.

Isolated - The local traffic signal controller operates in vehicle actuated mode with no coordination between sites. There is no defined cycle time. Each phase has a minimum and maximum green time. After the minimum green time has occurred, the controller will stay in the current phase until either the maximum green time is reached or the detector loops record no more vehicles using the phase green time. The phase then “gaps out” and the controller moves to the next demanded phase in a predefined phase sequence.

Master-isolated - Similar to Masterlink in that the cycle time and all phase lengths are calculated by the SCATS software. The only difference between Masterlink and Master-isolated is that there is no defined coordination point to be maintained for coordination purposes. This means that all phases can “gap” out.

Masterlink - The local traffic signal controller receives instructions from the SCATS regional computer. Cycle time, phase splits, and coordination are determined by the SCATS software. A requirement for Masterlink operation is that one of the operating phases must be permanently demanded never be

Project 232984 9 October 2013 Revision 2 Page 3

allowed to terminate early and it must use up any time within the cycle not used by the other phases. This requirement ensures that coordination is maintained.

Phase - A group of movements that are setup with SCATS to run in conjunction with each other.

SCATS - Sydney Coordinated Adaptive Traffic System, the signal control system used in WA. It monitors traffic signals and reacts to vehicles detections by changing the signal cycle time, phase durations, and coordination between signals.

Single diamond overlap - A signal phase setup for a four leg intersection where the opposing right turn movements on the primary road run in conjunction with each other, followed then by the primary road through movements. The option exists to run only one right turn with the adjacent through movement if there are no right turn vehicles on one of the approaches.

Split Approach - When one approach runs its own phase, once finished the opposing approach runs its phase.

Project 232984 9 October 2013 Revision 2 Page 4

1. Introduction

1.1 Project purpose Of all traffic management measures, the operation of traffic signals is often considered to have the most influence on the performance of metropolitan road networks. A well set up and maintained traffic signals system ensures that the maximum possible performance and capacity is achieved from existing network infrastructure. Likewise poorly performing intersections can unnecessarily constrain traffic flows, induce bottlenecks and can be unsafe.

RAC WA commissioned Aurecon to undertake an analytical and qualitative assessment of traffic signals operation and planning within the context of the Perth metropolitan road network. Main Roads WA was a key partner in the project. The aim of the analytical assessment is to highlight elements where signals are working well and similarly where traffic signals processes could be improved for the benefit of all road users. This included an assessment of the operational performance of selected demonstration sites, which included consideration of efficiency, road safety and the broader user experience.

1.2 Approach The project was delivered in two phases. The first phase focused on the initial investigations of several sites, looking at the signal operation, surrounding land use, intersection geometry and facilities for other road users particularly relating to safety. The intent was to agree and confirm the sites most suitable for further investigation.

Phase two primarily involved analysis of the performance of the intersections and corridors, using micro-simulation modelling with integrated Sydney Coordinated Adaptive Traffic System (SCATS) signal control. This predicted how the sites were performing and was used as a tool to test a range of improvement options.

This document reports on the results of the alternate approach to refining signal phasing and timing and makes recommendations regarding signal operation and road geometry, giving due consideration to all road users and road safety outcomes.

Project 232984 9 October 2013 Revision 2 Page 5

2. Perth Context

2.1 Governance The day to day operational maintenance of approximately 900 traffic signals is performed by Main Roads WA. The Main Roads WA Traffic Operations Centre (TOC) monitors traffic signal operation on all roads in WA using several systems including a network of approximately 200 CCTV cameras. TOC operators detect incidents on arterials, manage traffic generated by events and are also responsible for the implementation of all new signalised intersections.

Other state authorities and local councils work with Main Roads WA in the planning and design of signals relating to urban development, changing land use, public transport infrastructure, and other such schemes. However, only Main Roads WA is responsible for the performance of the road network once signals are operational.

2.2 Traffic signals system All road traffic signals within Perth and Western Australia (WA) operate using the SCATS signal system.

SCATS monitors traffic volumes in real-time and coordinates traffic signals to ease traffic congestion and improve traffic flow. SCATS was originally developed by the New South Wales Roads and Maritime Services and is now used in all Australian States to some extent.

SCATS operates by adjusting the coordination and green time of signals depending on the traffic volumes detected on each approach. The coordination and green time is generally adjusted to favour arterial corridors whilst also balancing the delay across the remaining approaches.

SCATS operates on three levels:

 Traffic Signal Controller: metal cabinets are located on the road verge near traffic lights which contain electronics and hardware (the traffic signal controller) that operates traffic signals. Sensors or vehicle detector loops are installed in the traffic lanes and these register and relay information on traffic flow or ‘demand’ on each arm of an intersection to the traffic signal controller. Although the signal controller is controlled by SCATS, it does have the ability to make decisions that affect local on-street operations. For example, the signal controller can use the information from the detector loops to decide whether to extend a signal phase or terminate the phase early if no traffic remains. The signal controller also sends the detector information back to the SCATS regional computer for further analysis.

 SCATS Regional Computers: There are several regional computers, each controlling up to 250 separate signalised intersections. Each regional computer combines the on-road information with pre-programmed data, which results in traffic signal operation and coordination requirements for its signalised intersections.

 SCATS Central Monitoring Computer: The regional computers are networked to a central monitoring computer located at the TOC. This computer enables operators to monitor traffic conditions and make adjustments to traffic signals operation.

Project 232984 9 October 2013 Revision 2 Page 6

3. Methodology

Objectives, policies, and constraints are very important factors for signal refinement practices. Policies and guidelines describe the agency’s approach and the strategies to be employed in signal adjustment relative to the prioritisation of objectives and resolving of conflicting objectives. This section sets out the methodology applied to test possible signal refinement options.

3.1 Stage One – Initial investigations

3.1.1 Agree and confirm methodology A discussion was held with Main Roads WA to present and agree the project objectives and approach and to understand the TOC’s systems including challenges and constraints as well as successes and opportunities.

3.1.2 Site selection Several potential demonstration sites where further efficiencies could be gained were identified using the expertise and knowledge of the project team and site visits undertaken during the AM, midday, PM peaks to confirm if the demonstration sites were suitable for further investigation.

During the site visits, observations were made about signal operation, land use, intersection geometry, pedestrian demand, and the extent of signals coordination with adjacent intersections and the information presented in a Stage one report.

Following the site visits, the project team agreed on two sites to be carried forward into stage two. The sites that were not selected were deemed either not suitable for this study or were assessed as currently operating efficiently as occurred in several cases.

These were:

 Site 1: Kelvin Road/ Tonkin Highway This intersection is commonly raised as an issue in correspondence from RAC members.

 Site 2: Orrong Road corridor A busy corridor (comprising Francisco Street, Archer Street, Wright Street and Oats Street intersections) with multiple land uses each placing their own demands on the four sets of signals.

3.1.3 Data collection Following confirmation of the two demonstration sites, the following data was collected:

 Count data: Austraffic was commissioned to collect full-day traffic count data including turning counts for each of the demonstration sites over two consecutive days (Tuesday and Wednesday) during normal flow conditions i.e. weeks that do not include a public holiday, school or university holidays.  SCATS data: SCATS input and output files and traffic signal plans.  Crash data: five year crash data was obtained from Main Roads WA.

3.1.4 Scan of other practice A scan of signals operation and emerging technology was undertaken for other cities and captured in a paper. This was used to inform a feature article prepared for the RAC’s Horizons Magazine on balancing the needs of various road users at signalised intersections.

Project 232984 9 October 2013 Revision 2 Page 7

3.2 Stage Two – SCATS options testing

3.2.1 Model build - existing situation Using the traffic counts and on-street traffic observations, models were built and calibrated for each of the base year AM, PM, midday and off peak time periods using the VISSIM micro-simulation traffic analysis tool with integrated SCATS signal control. These models replicated the existing situation at each of the sites.

The model build methodology is key to the robustness of the model and its results. This modelling methodology used a range of data sources to form a solid foundation for the option testing. The steps were as follows:

1. Carry out site visits of the demonstration sites to observe traffic patterns and driver behaviour. 2. Build network using GIS aerial imagery with data such as signal setups coded in. 3. Check and balance turn count data to ensure anomalies between count sites were minimised and that latent demand was considered. 4. Distribute traffic on the corridor using the ‘higher tier model cordon 2011 matrices’ from the Main Roads WA Regional Operations Model (ROM). 5. Establish wider network demand by placing the ROM distribution through a Furness method to align it with the turn count data.

3.2.2 Model build - option testing A workshop was held with Main Roads WA, RAC WA, and Aurecon to discuss and develop option testing plans for the demonstration sites. Options regarding improvements to the signal operations were identified for testing. These options consisted of technical changes to the signal operation and their 2013 model year performance was then compared to the existing on-street situation in terms of changes in delay and average speed. From this, the options that were performing closest to the optimum level could be identified.

3.2.3 Main Roads WA TOC signal adjustment process A third workshop/meeting was held with Main Roads WA, RAC WA, and Aurecon to discuss the process of SCATS signal adjustment and resourcing within the Main Roads WA TOC. This meeting adopted a strengths, weaknesses, opportunities, and threats (SWOT) assessment approach and also considered what would ideally occur if there were fewer constraints and what occurs in some TOCs in other Australian States, looking at:

 Objectives in signal refinement  Techniques used when refining signals  Time intervals between signal audits  Signal audit processes  Constraints in signal refinement  TOC resourcing.

Project 232984 9 October 2013 Revision 2 Page 8

4. Site Overviews

The following two sites were selected for this study due to their range of competing traffic movements, land uses, safety considerations, and for being commonly raised as an issue in correspondence from RAC members.

4.1 Tonkin Highway/ Kelvin Road

4.1.1 Location and geometry The intersection of Tonkin Highway/ Kelvin Road is located in the suburb of Orange Grove, approximately 18 kilometres south east of the Perth CBD (refer to Figure 4-1 and Figure 4-2).

Figure 4-1: Tonkin Highway/ Kelvin Road site location map

Figure 4-2: Tonkin Highway/ Kelvin Road site location map

Project 232984 9 October 2013 Revision 2 Page 9

The intersection of Tonkin Road and Kelvin Road is a signalised four arm intersection. The layout, as shown in Figure 4-3, consists of multiple lanes for each approach with signalised and non-signalised left turn slip lanes.

Figure 4-3: Tonkin Highway/ Kelvin Road intersection layout

The Main Roads WA classifications state that Tonkin Highway is a Primary Distributor with generally a 100 km/h speed limit, but this is reduced down to an 80 km/h speed limit at the Kelvin Road intersection.

Kelvin Road is a ‘Distributor A’ road on the south western approach and a ‘Distributor B’ road on the north eastern approach with a 70 km/h speed limit on both approaches.

4.1.2 Land use and subsequent traffic patterns While the site is within the Perth metropolitan area, there is no proximate land use apparent apart from some light industrial land plots. Traffic patterns are therefore predominantly formed by vehicles passing through from the outer Perth metropolitan suburbs heading north in the morning period and south in the evening period. This flow is combined with heavy commercial vehicle movements.

4.1.3 Signal operation The signals at Tonkin Highway/ Kelvin Road operate in a single diamond, split approach arrangement, i.e. the Tonkin Highway right turn movements run first, followed by the through movements, and then each of the Kelvin Road approaches run individually. Due to heavy demand, the right turn from the Tonkin Highway northwest approach often gets a call back to run twice within a cycle. High cycle times exist in an attempt to minimise the number of stops for drivers.

4.1.4 Crash history Crash data from Main Roads WA’s Crash Analysis Reporting System (CARS) was analysed for the Tonkin Highway/ Kelvin Road intersection. Further crash data analysis is attached in Appendix A.

Project 232984 9 October 2013 Revision 2 Page 10

A total of 203 crashes, involving all modes of transport, occurred over the 2008 – 2012 period at the Tonkin Highway/ Kelvin Road intersection. Table 4-1 below shows that with the exception of a low number of crashes in 2010, the total crash numbers has been reasonably consistent between years.

Table 4-1: Total number of crashes 2008-2012

60

50

40

30

20

10

0 2008 2009 2010 2011 2012

The crash data at the Tonkin Highway/ Kelvin Road intersection was also compared to other intersections within the City of Gosnells and the State. According to the Main Roads WA Intersection Crash Ranking Interactive Report, which covers all intersections which have had one or more reported road crashes over 5 years in the period ending 31 December 2012, the intersection was ranked as follows:

 State frequency rank no. 123, with 103 crashes within the given time period. The intersection of Eelup Rotary and Perth-Bunbury Highway is the worst ranked intersection in WA with 605 crashes.  State cost rank no. 40, with a total crash cost of $6,670,493. The intersection of Tonkin Highway/ Horrie Miller Drive is the worst ranked intersection in WA with a crash cost of $19,035,513.  Within the City of Gosnells, the intersection ranks at no. 7 in terms of crash frequency. The intersection at Nicholson Road / Yale Road is the worst ranked intersection within the City of Gosnells with 305 crashes.  Within the City of Gosnells, the intersection ranks at no.3 in terms of crash cost. The intersection at Nicholson Road / Yale Road is the worst ranked intersection within the City of Gosnells with a crash cost of $8,100,303.

While no specific aspects from the crash data was tested in the model option tests, any reductions in congestion, such as those achieved by refining the signal settings, often leads to a reduction in crashes due to less stop/start vehicle actions occurring.

4.1.5 Active transport demand With no immediate residential or other land use activities nearby, active transport demand is low and is estimated to be less than one pedestrian or cyclist overall per cycle. The pedestrian facilities are

Project 232984 9 October 2013 Revision 2 Page 11

missing line markings and drop down kerbs and some of the pedestrian lanterns do not have symbolic displays.

4.1.6 Key observations The on-site observations of traffic patterns were as follows:

Morning Peak Period, 06:30 – 07:30

 Heavy queuing for the Tonkin Highway southeast approach through movement, beyond sightline of curve, approximately 800m.  Tonkin Highway northwest approach right turn bay fills, sometimes blocks through lane to extend back towards 100 km/h area. This creates a safety issue with an increased risk of rear end crashes, especially considering the curvature of Tonkin Highway in that section of road.  High left turn volume for the Kelvin Road southwest approach.

Midday Period, 12:00 – 13:00

 Tonkin Highway southeast approach through movement has a sizeable queue.  Tonkin Highway northwest approach right turn bay intermittently filling.

Evening Peak Period, 16:00 – 17:00

 Tonkin Highway northwest approach through movement queuing goes beyond the corner intermittently, approximately 400m.  Tonkin Highway northwest approach right turn bay is close to being full.  Tonkin Highway southeast approach through movement and Kelvin Rd west approach queues are approximately 100m in length.

Off-Peak Period, 21:00 – 22:00

 Low volumes.

4.2 Orrong Road

4.2.1 Location and geometry The Orrong Road study corridor is 2.2 kilometres long and located approximately 4.5 kilometres south east of the Perth CBD (refer to Figure 4-4 and Figure 4-5).

Project 232984 9 October 2013 Revision 2 Page 12

Figure 4-4: Orrong Road looking northwest to Oats Street

Figure 4-5: Orrong Road site location map

The section of the Orrong Road corridor assessed in this study includes four signalised intersections. The intersection layouts are shown in Figure 4-6 to Figure 4-9:

Project 232984 9 October 2013 Revision 2 Page 13

 Orrong Road/ Francisco Street

Figure 4-6: Orrong Road/ Francisco Street intersection layout

 Orrong Road/ Archer Street

Figure 4-7: Orrong Road/ Archer Street intersection layout

Project 232984 9 October 2013 Revision 2 Page 14

 Orrong Road/ Wright Street

Figure 4-8: Orrong Road/ Wright Street intersection layout

 Orrong Road/ Oats Street

Figure 4-9: Orrong Road/ Oats Street intersection layout

Project 232984 9 October 2013 Revision 2 Page 15

The Main Roads WA classifications state that Orrong Road is a ‘Primary Distributor’ with a 70 km/h speed limit. The southwest approach of Oats Street is a ‘Distributor A’ road, Archer Street is a ‘Distributor B’ road, and Francisco Road, Wright Street, and the northeast approach of Oats Street are ‘Local Distributors’. All of these side roads have 50 km/h speed limits.

4.2.2 Land use and subsequent traffic patterns The Orrong Road site is impacted by various land use types. The site is immediately surrounded by low density residential housing that has access to Orrong Road both directly and via the side roads. Orrong Road also provides one of the main links into the Perth Central Business District (CBD) from the east and as such has high eastbound and westbound traffic flows in the morning and evening peak periods respectively. A large industrial area exists to the east of the study corridor, resulting in high traffic flows in the counter direction. Carlisle Primary School is located at the corner of Orrong Road and Wright Street, increasing pedestrian demand at this location.

4.2.3 Signal operation The four signalised intersections along the corridor are spaced over 2.2 km of road and they operate in a coordinated fashion however it is believed that this can be further improved. The two middle signalised intersections are T-intersections with the remaining two being four arm intersections. The levels of service at the signalised intersections seem to be unbalanced with the Orrong Road corridor not favoured enough compared to the side roads.

Intersection signal operations are as follows:

 Orrong Road/ Francisco Street: Single Diamond overlap with split side roads, i.e. the right turn movements from Orrong Road run first, followed by the Orrong road through movements and lastly the Francisco Street approaches running individually.  Orrong Road/ Archer Street: Standard Three Phase Tee with a lead right turn into Archer Street, i.e. the Orrong Road west approach movements run first followed by the Orrong Road through movements, with the last movements being from Archer Street.  Orrong Road/ Wright Street: Four Phase Repeat Right Turn with a pedestrian crossing on both sides of the stem of the Tee, i.e. the Orrong Road west approach movements runs first followed by the Orrong Road through movements, then Orrong Road west approach movements again, and lastly the Wright Street movements.  Orrong Road/ Oats Street: Single Diamond Overlap plus a right turn phase off Oats Street southwest approach, i.e. the right turn movements from Orrong Road run first, followed by the Orrong Road through movements, then the Oats Street south approach movements, and lastly all Oats Street movements with the right turn movements giving way to opposing traffic.

Project 232984 9 October 2013 Revision 2 Page 16

4.2.4 Crash history Crash data from CARS was analysed for the Orrong Road corridor between Francisco Street and Oats Street. Further crash data analysis is attached in Appendix A.

A total of 1,493 crashes, involving all modes of transport, occurred over the 2008 – 2012 period (refer to Figure 4-10).

Francisco St / Francisco Pl Alexander Rd Wright St Fulham St

Mercury St Oats St Roberts Rd Archer St Signalised Intersection Figure 4-10: Orrong Road corridor (Francisco Street to Oats Street)

There are four signalised intersections along the corridor including Francisco Street, Archer Street, Wright Street and Oats Street.

Not surprisingly, 77% of crashes along the corridor occurred at an intersection and nearly 70% of these crashes occurred at one of the four signalised intersections. The number of crashes at the signalised intersections is shown in Table 4-2. The highest number of crashes at a signalised intersection occurred at Orrong Road/ Oats Street and the worst midblock segment was between Archer Road and Wright Street.

Table 4-2: Crash locations at signalised intersections

Location of Crash Number of Crashes

Orrong Rd / Francisco St intersection 202

Orrong Rd / Archer St intersection 185

Orrong Rd / Wright St intersection 132

Orrong Rd / Oats St intersection 274

Table 4-3 shows that the number of crashes across the entire corridor has a pattern of slowly increasing, likely to be partly due to increasing traffic volumes.

Project 232984 9 October 2013 Revision 2 Page 17

Table 4-3: Total number of crashes from 2008-2012

400

350

300

250

200

150

100

50

0 2008 2009 2010 2011 2012

The crash statistics presented in Appendix A show the nature of the crash event for the particular vehicle movement. From this, it was identified that the prevailing crash movement is in the east-west and west-east direction, (i.e. on Orrong Road) with the highest proportion of these being rear end crashes.

The crash data for the intersections in the Orrong Road corridor was also compared to other intersections within the Town of Victoria Park and the State. According to the Main Roads WA Intersection Crash Ranking Interactive Report, which covers all intersections which have had one or more reported road crashes over 5 years in the period ending 31 December 2012, the following was summarised:

Table 4-4: Crash ranking comparison for Orrong Road corridor

Intersection State Frequency State Cost Rank Frequency Rank Cost Rank within Rank within Town of Town of Victoria Victoria Park Park

Orrong Road / 123 160 3 4 Francisco Street

Orrong Road / 153 157 4 3 Archer Street

Orrong Road / 287 500 11 15 Wright Street

Orrong Road / 61 126 1 2 Oats Street

As with the Tonkin Highway/ Kelvin Road demonstration site, no specific aspects from the crash data were tested in the model option tests.

Project 232984 9 October 2013 Revision 2 Page 18

4.2.5 Active transport demand With the surrounding residential land use, active transport demand is at a moderate level, with the Carlisle Primary School at the corner of Orrong Road and Wright Street contributing to this active transport demand. Pedestrian facilities are well marked and well provided however there are no dedicated cyclist facilities. The high cycle times result in long wait times for pedestrians.

4.2.6 Key observations The on-site observations of traffic patterns were as follows with journey times collected on Orrong Road from Leach Highway to the Great Eastern Highway overpass.

Morning Peak Period, 07:00 - 08:00

 Large levels of congestion for the westbound through movement, coordination can be improved but would not be done through increasing cycle times.  Westbound queue extends back from Oats Road to nearly the Leach Highway Interchange.  Eastbound travel time between 5 to 6 minutes, westbound travel time between 5 to 12 minutes.

Midday Period, 12:00 - 13:00

 Steady traffic flows with short queues.

Evening Peak Period, 16:30 - 17:30

 Large levels of congestion for both the eastbound and westbound directions of travel.  Eastbound queue extending back from Francisco Street to the Great Eastern Highway bridge.  Westbound queue extending back from Oats Road to nearly the Leach Highway Interchange.  Eastbound travel time between 5 to 8 minutes, westbound travel time between 6 to 8 minutes.

Off-Peak Period, 21:00 - 22:00

 Low volumes.

Project 232984 9 October 2013 Revision 2 Page 19

5. Model Build and Calibration: Existing Situation

5.1 Overview The modelling phase of the project involved the analysis of the performance of the sites using micro- simulation modelling with integrated SCATS signal control. The model provides a base to which the impact of tested options could be quantitatively compared.

The first stage of model build was the development of calibrated and validated base VISSIM models, which reflected existing traffic conditions at each intersection. These base models were then adjusted to create new models to test each proposed option. By minimising uncertainty around impacts, these assessments provide valuable information that can be used to aid decision making regarding the future operation of the demonstration sites.

5.1.1 Tonkin Highway/ Kelvin Road model area The micro-simulation model was developed for the Tonkin Highway/ Kelvin Road intersection and its respective approaches, as shown in Figure 5-1.

Figure 5-1: Model area for Tonkin Highway/Kelvin Road

Tonkin Highway Kelvin Road

Legend

Signalised Intersection

Project 232984 9 October 2013 Revision 2 Page 20

5.1.2 Orrong Road corridor model area The Orrong Road corridor model covers an area from north-west of Francisco Street to south-east of Leach Highway and major intersections in between. While the model build covers several intersections, the model calibration (Section 5.4) was only focussed on the four key signalised intersections, as described in Section 4.2. The Leach Highway/ Orrong Road Interchange was included in the model area so that its links with the SCATS signal corridor could be considered and to incorporate the platooning traffic patterns along the Orrong Road corridor that the interchange creates whereby traffic moves along the corridor in a cluster. The model area is shown in Figure 5-2 below.

Figure 5-2: Model area for the Orrong Road corridor

Orrong Road

Francisco Street

Alexander Road

Roberts Road

Wright Street Archer Street

Oats Street

Oats Street

Leach Highway

Legend

Signalised Intersection

Project 232984 9 October 2013 Revision 2 Page 21

5.2 Data sources Calibration of models to existing traffic conditions is only as effective as the quality of data used to build and calibrate the model. For this project several sets of data were collected and therefore a robust amount of data has been used. This data is detailed below.

Gathered data:

 Classified turning counts (by Austraffic) for each of the sites collected on Tuesday the 9th and Wednesday the 10th of April 2013, from 06:00 to 22:00  Used to ascertain the traffic demands and calibrate the model to.  On-site journey time data collection  Used for the validation of the model  RAC WA journey time data, 2011  Used as a check for intersection to intersection travel times  Site aerial imagery from GIS  Used to code the modelled network links to  SCATS input files including the LX, ram, TC, sft, and central manager database files  Used for the SCATS signal control within the model.  SCATS output files including the VS, SM, and IDM files  Used to review the on-street signal performance and to provide realistic pedestrian demands at each signalised intersection.  Signal plans for each selected intersection  Used to check the signal phasing and setup.  ROM cordon matrices  Used to establish the distribution of traffic within the network.

5.3 Assumptions A model is essentially formed by developing a set of assumptions, some regarding the model inputs and operations, and others which are prebuilt into the modelling software. Some of the key assumptions that apply to this study are:

 Only signalised intersections and major side roads have been included in the models.  The survey data used to build the traffic demands replicates an average day.  SCATS input files and associated signal settings from the provided files are the same as those used on site.  The ROM model distribution has been adjusted in line with the traffic counts during the calibration process.  Vehicle spacing and average vehicle lengths have been adjusted to better replicate current WA conditions. Vehicle spacing is 2.3m and vehicle lengths upper range is up to 5.17m.  No changes in traffic volumes occurred between the current base and option models.

5.3.1 Tonkin Highway/ Kelvin Road peak hour periods A range of data was collected and in most instances three hour models were developed with the primary peak hour being calibrated, validated, and analysed for. The Tonkin Highway/ Kelvin Road peak hours were established from the turn count data and were as follows:

Project 232984 9 October 2013 Revision 2 Page 22

 AM peak period: 0645 to 0745  Midday peak period: 1145 to 1245  PM peak period: 1600 to 1700  Off-peak period: 2100 to 2200

5.3.2 Orrong Road peak hour periods The Orrong Road corridor peak hours were found to be as follows:

 AM peak period: 0715 to 0815  Midday peak period: 1130 to 1230  PM peak period: 1615 to 1715  Off-peak period: 2100 to 2200

5.4 Model calibration and validation Model calibration and validation is necessary to ensure that a model accurately represents an existing traffic situation and can be used with confidence to test alternatives. Model calibration is the process of comparing observed data to the model output data to check that the model reflects on-street conditions. Calibration of the model has been based on the following:

 Vehicle Behaviour: Undertaking a visual check to confirm the observed vehicle behaviour in the model is consistent with that observed on-street.  Turn Counts: Comparing observed and modelled turning movements for general traffic over the modelled peak hour periods.  Link Counts: Comparing observed and modelled link counts for general traffic over the modelled peak hour periods.  Queue Lengths: Undertaking a visual check to confirm the modelled queue operation is consistent with those observed on site.  Signal timings: Undertaking a check to confirm that the observed signal timings in the model are relatively consistent with those collected on-street.

After the model is calibrated, validation is undertaken. Validation is the process of comparing the model outputs against independently measured data, which was not used in the calibration process, to confirm the model replicates existing network conditions. For the purposes of this study, validation is based on comparisons of observed and modelled journey travel times for general traffic over the modelled peak hour periods.

While Roberts Road, Alexander Road, and Leach Highway are included in the model, their volumes are only indicative and have not been calibrated. No analysis has been conducted at these intersections.

5.4.1 Tonkin Highway/ Kelvin Road turn and link count calibration

A Traffic Modelling Guidelines document is made available by the New South Wales Roads and Maritime Services (NSW RMS). As a summary the network wide calibration criteria from this document have been used for this study as well as the more useable criteria issued through the New Zealand Economic Evaluation Manual (NZTA EEM). These guidelines are used in the absence of any similar WA documents currently available.

Project 232984 9 October 2013 Revision 2 Page 23

As shown below, all of the turn and link count calibration criteria have been met as would be expected with a single intersection model and therefore the model is deemed calibrated. Further details regarding the calibration are contained in Appendix B.

Table 5-1: NSW RMS Criteria for Turn Counts at Tonkin Highway/ Kelvin Road Criteria AM Midday PM Off-peak GEH < 5 Minimum 85 per cent 100% 100% 100% 100% Counts with GEH > 10 require explanation 0 0 0 0 R2 to be > 0.9 0.9967 0.9998 0.9990 0.9997

Table 5-2: NZTA EEM Criteria at Tonkin Highway/ Kelvin Road

Criteria and Measures Range AM Midday PM Off-peak < 99 vph 83% 100% 100% 90% 100 -199 vph 100% - 100% - Calibration Acceptance Targets - Within 200 - 499 vph 100% 100% 100% 100% 30% of Observed 500 - 999 vph 100% 100% 100% - > 1,000 vph 100% - 100% - > 100 vph 100% 100% 100% 100% Turn Criteria and Measures Criteria GEH statistic < 5.0 for individual turn >60% of cases 100% 100% 100% 100% volumes GEH statistic < 10.0 for individual turn >95% of cases 100% 100% 100% 100% volumes GEH statistic < 12.0 for individual turn 100% of cases 100% 100% 100% 100% volumes Link Count R2 Measure Criteria R2 value for modelled versus observed > 0.85 0.9958 0.9997 0.9988 0.9999 volumes for all individual Links Links Criteria and Measures Criteria GEH statistic < 5.0 for individual Links >60% of cases 100% 100% 100% 100% volumes GEH statistic < 10.0 for individual Links >95% of cases 100% 100% 100% 100% volumes GEH statistic < 12.0 for individual Links 100% of cases 100% 100% 100% 100% volumes Criteria and Measures Criteria Percentage RMSE Turns <30% 13% 3% 7% 4%

5.4.2 Tonkin Highway/ Kelvin Road queue length calibration Queuing is an inherently unstable phenomenon which can vary greatly from day to day. Queue measurements can be very subjective as the definition of what vehicles count as “queued” can differ between observers and between modelling packages.

Notwithstanding the issues relating to queue variability, queue comparisons are still valuable and have been made between observed and modelled queue length data to ensure queue patterns are appropriately represented in the model.

In the AM period substantial queuing exists northbound on Tonkin Highway and stretches several hundred metres. Conversely there is queuing on the Tonkin Highway southbound right turn that often exceeds its turn bay length as well as on the Kelvin Road western approach for the left turn movement. In the PM period queuing is not as pronounced but is still present on Tonkin Highway in the northbound and southbound directions.

Project 232984 9 October 2013 Revision 2 Page 24

These observed queues have been replicated in the model.

5.4.3 Orrong Road turn and link count calibration Calibration results across the four selected signalised intersections on the Orrong Road corridor are shown in Table 5-3 and Table 5-4. All of the turn and link count calibration criteria has been met and the model can therefore be deemed calibrated.

Table 5-3: NSW RMS Criteria for Turn Counts for the Orrong Road Corridor Criteria AM Midday PM Offpeak GEH < 5 Minimum 85 per cent 94% 100% 97% 97% Counts with GEH > 10 require explanation 0 0 0 0 R2 to be > 0.9 0.9901 0.9966 0.9964 0.9937

Table 5-4: NZTA EEM Criteria for the Orrong Road Corridor

Criteria and Measures Range AM Midday PM Offpeak < 99 vph 60% 92% 78% 80% 100 -199 vph 100% 100% 100% 100% Calibration Acceptance Targets - Within 200 - 499 vph 86% 100% 89% 100% 30% of Observed 500 - 999 vph 100% - - 100% > 1,000 vph 100% 100% 100% - > 100 vph 95% 100% 96% 100% Turn Criteria and Measures Criteria GEH statistic < 5.0 for individual turn >60% of cases 94% 100% 97% 97% volumes GEH statistic < 10.0 for individual turn >95% of cases 100% 100% 100% 100% volumes GEH statistic < 12.0 for individual turn 100% of cases 100% 100% 100% 100% volumes Link Count R2 Measure Criteria R2 value for modelled versus observed > 0.85 0.9814 0.9929 0.9913 0.9872 volumes for all individual Links Links Criteria and Measures Criteria GEH statistic < 5.0 for individual Links >60% of cases 89% 100% 93% 96% volumes GEH statistic < 10.0 for individual Links >95% of cases 100% 100% 100% 100% volumes GEH statistic < 12.0 for individual Links 100% of cases 100% 100% 100% 100% volumes Criteria and Measures Criteria Percentage RMSE Turns <30% 14% 8% 9% 8%

5.4.4 Orrong Road queue length calibration In the AM period the predominant queues are westbound, primarily at Oats Street but also eastbound at Archer Street. The PM period again has some queuing westbound on Orrong Road but the primary queue is late in the PM period where traffic heading eastbound from the Graham Farmer Freeway arrives at the Orrong Road corridor and more specifically the Francisco Street intersection. These observed queues have been replicated in the model.

Project 232984 9 October 2013 Revision 2 Page 25

5.4.5 Orrong Road journey time validation Journey time data was collected during the peak AM and PM hours by Aurecon using a floating car method on a day of the Austraffic surveys. The journey time data was recorded for both the eastbound and westbound directions from after the Orrong Road/ Leach Highway Interchange to the Great Eastern Highway over pass.

As well as this set of data, the RAC WA provided a further 2011 journey time data collected between individual intersections. Due the collection of the RAC WA data preceding the study a direct comparison could not be made, rather the data has been referred to. These journey times have also been recorded over a different distance to those used in the RAC WA surveys and those used in the option model comparisons further in the report.

The NSW RMS Modelling Guidelines suggest modelled journey times should be within 15% or 1 minute of the observed value, whichever is greater.

The eastbound and westbound journey time comparisons are shown below. All of the journey time routes match the set criteria with the exception of the AM period eastbound route, which is 1:11 over the observed value. However, only two observed runs form the basis of this comparison, and referring to the RAC WA data for a shorter route, journey times vary from 4:52 minutes to 7:11 minutes for a similar time period. On this basis, the modelled journey time has been deemed acceptable.

Table 5-5: Westbound Journey time comparison for the Orrong Road corridor Journey Time Source AM PM Average Observed Journey Time 9:32 6:35 Average Modelled Journey Time 8:15 6:06 Difference (min:secs) 1:17 0:29 % Difference -13.5% -7.3%

Table 5-6: Eastbound Journey time comparison for the Orrong Road corridor Journey Time Source AM PM Average Observed Journey Time 5:37 5:42 Average Modelled Journey Time 6:48 6:05 Difference (min:secs) 1:11 0:23 % Difference 21.1% 6.7%

Project 232984 9 October 2013 Revision 2 Page 26

6. Option Testing: Tonkin Highway/ Kelvin Road

6.1 Option plan A workshop was held in order to present a series of options that were proposed for testing to Main Roads WA. These options were agreed by RAC WA and Main Roads WA and the following option test plan was developed.

Option Option Details

Option 1 Adjust SCATS signal settings including Master-isolated operation (maximum green times for each phase are determined by SCATS based on the approach degree of saturation) and optimal cycle time, no infrastructure costs. Option 2 Test under a Double diamond overlap phasing, minor infrastructure costs in the order of $1,000’s. Option 3 Double right turn on Tonkin Highway north approach and localised additional lane flaring on the Tonkin Highway south approach, medium infrastructure in the order of $100,000’s. Option 4 Ultimate test, adjustment of the SCATS signal settings including Master-isolated operation, optimal cycle time, and test under a Double diamond overlap phasing, minor infrastructure costs in the order of $1,000’s.

6.2 Option testing results All of the result tables are presented with shades of green or red to represent the relative positive or negative impacts of the option tests. Focus has been given to delay and level of service as this best evaluates changes in performance of an intersection. The total columns take into account the cumulative effect of the traffic volumes over all of the periods.

6.2.1 Delay The tables below present the total delay of all the vehicles in the network recorded for the base model and the subsequent option models for all four of the modelled time periods. These tables show that as a total Option 3 achieves up to a 36% reduction in delay over the base model for all of the model periods. Options 1, 2, and 4 also show a reduction in delay and even though these reductions are to a lesser extent, these options require less infrastructure works than Option 3.

Using the following formula the annual delay cost was calculated based on 260 travel days (workdays) per year and an average hourly rate of AUD$35 and vehicle occupancy of 1.2.

Calculation: Daily Hours of Reduction in Vehicle Delay x Vehicle Occupancy = Daily Person Hour Delay Then: Daily Person Hours Delay x 260 Days (work week) x Cost per Hour = Annual Delay Savings

Equating this to annual delay cost to occupants, Option 4 would save approximately $802,000 per year in lost time cost.

Project 232984 9 October 2013 Revision 2 Page 27

Table 6-1: Tonkin Highway/ Kelvin Road total vehicle delay (hours) AM Midday PM Off-Peak Total Base 139.8 23.9 106.3 4.3 274.3 Option 1 132.3 18.6 112.3 4.1 267.4 Option 2 124.6 19.9 100.1 3.3 248.0 Option 3 58.8 21.5 92.5 4.3 177.0 Option 4 91.9 16.6 89.1 3.2 200.8

Table 6-2: Tonkin Highway/ Kelvin Road total vehicle delay (percentage change from base) AM Midday PM Off-Peak Total Option 1 -5.4% -22.2% 5.7% -4.1% -2.5% Option 2 -10.8% -16.9% -5.8% -22.9% -9.6% Option 3 -57.9% -10.1% -13.0% -0.5% -35.5% Option 4 -34.3% -30.5% -16.2% -24.3% -26.8% A breakdown of the delay for each movement and the associated levels of service are attached in Appendix C.

6.2.2 Average speeds An average speed comparison across the base and four option model networks was undertaken and is displayed in Table 6-3. The weighted total average speed values were calculated by multiplying the network volumes in each period by the average speeds in each period. These values were summed for all periods and divided by the total of all volumes to get the weighted total average speed for each option.

Calculation:

(AM volume x AM speeds) + (Midday volume x Midday speeds) + (PM volume x PM speeds) + (Offpeak volume x Offpeak speeds) = Total volume speed value

Then:

Total volume speed value / (AM volume + Midday volume + PM volume + Offpeak volume) = Weighted total average speed

By referring to the average speeds across all the models, Option 3 and Option 4 show a substantial increase in average speeds across all model periods with the exception of a reduction in the average speed for the off-peak period of Option 3.

Table 6-3: Tonkin Highway/Kelvin Road average speed (km/h) AM Midday PM Off-Peak Total Base 29.3 57.8 39.3 72.3 42.9 Option 1 30.0 62.3 37.1 72.6 43.4 Option 2 30.2 62.2 38.5 75.7 44.2 Option 3 45.6 58.2 40.8 71.6 48.7 Option 4 37.6 64.4 42.7 76.0 48.7

Project 232984 9 October 2013 Revision 2 Page 28

6.2.3 Carbon emissions The total estimated carbon emissions for all movements in the models were calculated within the VISSIM micro-simulation software and are displayed in Table 7-5 in the unit of grams. Option 3 has a significant reduction in carbon emissions across all model periods. It is understood that different vehicle operations in the model with the lower cycle times is the reason for higher carbon emissions for many of the options, despite having a reduction in the vehicle delays.

Table 6-4: Tonkin Highway/Kelvin Road carbon emissions (grams) AM Midday PM Off-Peak Total Base 6,986 4,597 8,949 1,397 21,927 Option 1 7,638 4,570 9,793 1,362 23,362 Option 2 7,644 4,593 9,752 1,270 23,259 Option 3 6,559 4,229 8,552 1,321 20,662 Option 4 7,796 4,290 8,653 1,264 22,003

6.3 Summary results By reviewing the total delays, average speeds, carbon emissions, and likely scale of cost for the civil works, Option 4 has been identified as the best performing option compared to the base model, most highlighted by its 26.8% reduction in delay over all the periods. The right turn from Tonkin Highway north into Kelvin Road westbound does have an increase in delay for Option 1 and Option 4 but a decrease in delay for Options 2 and 3. This increase in delay for the Option 1 and Option 4 north approach right turn is due to no call back running. It is anticipated that a civil works capacity upgrade for the north approach right turn bay is required regardless of the signal operation due to its impact on road safety.

Option 3 performs well but it has high cost civil works associated with it. Option 1 and Option 2 have overall positive increases in delay.

The increases in performance of the Tonkin Highway/ Kelvin Road intersection are achieved by implementing the following:

 A double diamond overlap phasing arrangement.  Master-isolated operation instead of flexi-link but this dependent on the treatment of the Tonkin Highway North approach right turn bay.  Lowering the current 250 second peak period cycle time.

Project 232984 9 October 2013 Revision 2 Page 29

7. Option Testing: Orrong Road

7.1 Option plan A workshop was held in order to present a series of options that were proposed for testing to Main Roads WA. Several of these options were agreed by RAC WA and Main Roads WA and the following option test plan was agreed.

Table 7-1: Option test plan

Option Option Details

Option 1 Determine the optimal cycle time and refine SCATS data which will include operating incremental split selection, no infrastructure costs. Option 2 Stagger the pedestrian crossing on Wright and Oat Streets, minor infrastructure costs in the order of $1,000s. Option 3 On south approach of the San Francisco Place implement a left turn only with a u-turn bay, on Archer Street and Wright Street implement a lag right turn. SCATS data refined. Minor infrastructure costs in the order of $10,000s. Option 4 Testing of Oats Street under a double diamond operation, minor infrastructure costs in the order of $10,000s. Option 5 Ultimate test, as per Option 1 with further refinement to SCATS data and coordination, no infrastructure costs.

7.2 Option testing results

7.2.1 Delay The tables below present the total vehicle delay (hours) for the base model and the subsequent option models for all four of the modelled time periods. No single option offers a reduction in delay over all of the model periods, however Option 5 shows the greatest overall reduction in delay.

Equating this to annual delay cost to occupants, Option 5 would save approximately $760,000 per year in lost time cost.

Table 7-2: Orrong Road Corridor total vehicle delay (hours) AM Midday PM Off-Peak Total Base 337.9 106.7 238.0 17.0 699.6 Option 1 278.3 134.2 243.2 15.8 671.6 Option 2 329.7 107.0 219.4 17.4 673.4 Option 3 284.2 109.2 258.3 15.7 667.4 Option 4 390.2 115.3 317.3 16.6 839.4 Option 5 281.2 126.3 206.6 15.9 629.9

Project 232984 9 October 2013 Revision 2 Page 30

Table 7-3: Orrong Road Corridor total vehicle delay (percentage change from base) AM Midday PM Off-Peak Total Option 1 -17.6% 25.8% 2.2% -7.0% -4.0% Option 2 -2.4% 0.3% -7.8% 2.1% -3.7% Option 3 -15.9% 2.4% 8.5% -7.8% -4.6% Option 4 15.5% 8.0% 33.3% -2.5% 20.0% Option 5 -16.8% 18.4% -13.2% -6.7% -10.0% A breakdown of the delay for each movement and the levels of service are attached in Appendix C. Note: there was some isolated queuing that at times extended beyond the model bounds but sensitivity tests showed that the effect of this on delay was minimal.

7.2.2 Average speeds An average speed comparison across all the links of the models was undertaken and is displayed in Table 7-4. The weighted total average speed values were calculated by multiplying the network volumes in each period by the average speeds in each period. These values were summed for all periods and divided by the total of all volumes to get the weighted total average speed for each option.

There is little overall change in average speeds compared to the base model however Option 2 and Option 5 are the best performing.

Table 7-4: Orrong Road average speeds (km/h) AM Midday PM Off-Peak Total Base 26.9 38.3 31.9 49.4 33.9 Option 1 29.1 36.9 30.2 50.0 33.8 Option 2 27.3 38.8 32.9 49.3 34.5 Option 3 28.4 39.4 29.8 49.9 34.1 Option 4 22.6 38.1 28.1 49.6 31.3 Option 5 28.7 37.9 33.2 49.8 34.8

7.2.3 Carbon emissions The total estimated carbon emissions for all movements in the models are displayed in Table 7-5 in the unit of grams. No model has a reduction in emissions over all the periods however Option 2 and Option 5 has a total reduction in carbon emissions.

Table 7-5: Orrong Road Corridor carbon emissions (grams) AM Midday PM Off-Peak Total Base 28,866 20,013 26,590 6,418 81,886 Option 1 26,786 20,953 27,641 6,038 81,419 Option 2 28,943 19,913 26,501 6,418 81,775 Option 3 27,273 19,776 28,161 6,044 81,255 Option 4 32,101 20,153 28,086 6,375 86,716 Option 5 27,546 20,484 26,614 6,092 80,736

7.2.4 Journey time comparison graphs Journey time comparisons, focusing on the Orrong Road corridor movements, are displayed below. They show that some marginal improvement is achieved with most options however some options perform poorly. In the AM period Options 2 and 3 show improvement in the journey times and in the

Project 232984 9 October 2013 Revision 2 Page 31

PM period Options 2 and 5 perform better than the base case. The largest improvement is experienced in the AM period eastbound by Options 1 and 5, with an average saving of over 2 minutes.

Figure 7-1: Eastbound Journey Times for the Orrong Road corridor in the AM peak hour period

Eastbound Journey Times - AM Peak 08:00

07:00

06:00

05:00 Base Option 1

(min:secs) 04:00 Option 2 03:00 Option 3 02:00 Option 4

TravelTime 01:00 Option 5

00:00 0 500 1000 1500 2000 2500 3000 3500 Distance (m)

Figure 7-2: Westbound Journey Times for the Orrong Road corridor in the AM peak hour period

Westbound Journey Times - AM Peak 10:00 09:00

08:00 07:00 Base 06:00 Option 1 05:00 Option 2 04:00 Option 3 03:00 Option 4 02:00

TravelTime(min:secs) Option 5 01:00 00:00 0 500 1000 1500 2000 2500 3000 3500 Distance (m)

Project 232984 9 October 2013 Revision 2 Page 32

Figure 7-3: Eastbound Journey Times for the Orrong Road corridor in the PM peak hour period

Eastbound Journey Times - PM Peak 07:00

06:00

05:00 Base

04:00 Option 1 (min:secs) 03:00 Option 2 Option 3 02:00 Option 4

TravelTime 01:00 Option 5

00:00 0 500 1000 1500 2000 2500 3000 3500 Distance (m)

Figure 7-4: Westbound Journey Times for the Orrong Road corridor in the PM peak hour period

Westbound Journey Times - PM Peak 09:00 08:00

07:00

06:00 Base

05:00 Option 1 (min:secs) 04:00 Option 2 03:00 Option 3 02:00 Option 4

TravelTime Option 5 01:00 00:00 0 500 1000 1500 2000 2500 3000 3500 Distance (m)

Project 232984 9 October 2013 Revision 2 Page 33

7.3 Summary results From this analysis Option 5 has been put forward as the recommended option due to its overall improvement over the base case scenario in terms of reducing delay by 10% and some improvement in the average speeds, carbon emissions, and Orrong Road journey times. There is also no civil work costs required for Option 5.

Option 5 achieved these improvements by implementing the following:

 A switch to incremental split selection.  Lowering the cycle time from the current range of 160-180 seconds.  Refinement of the corridor coordination. The alternative options only showed a marginal overall improvement in delay with the exception of the Oats Street double diamond overlap Option 4 which performs very poorly.

Project 232984 9 October 2013 Revision 2 Page 34

8. Key Findings and Conclusions

8.1.1 SCATS settings From this study of the SCATS signal operations and settings, several noteworthy aspects have emerged. Firstly, Main Roads WA currently has many of the demonstration sites in this particular study operating close to their optimal level. It is obtaining the final 5-10% of capacity where the difficulty lies, as shown with the Orrong Road option testing, and while some tested options improved the network performance, equally some of the other options had a detrimental effect on the network operation. By using micro-simulation modelling, this final level of optimisation can be attempted without the risk of a poorly performing option being tested on-street. This is especially important considering that most of the proposed measures should only be implemented on a site by site basis rather than broadly across the network.

The option testing highlighted that the following changes to the way SCATS is used would have a positive impact on the selected sites.

Cycle times A key finding is that reducing cycle times reduces the delay to road users. The Tonkin Highway/ Kelvin Road option testing found that lowering the cycle times in the peak periods from 250 seconds to 190 seconds, or preferably further to 145 seconds, reduced vehicle delay as well as the delay for any pedestrians waiting to cross at the signalised intersection. This final cycle time is dependent on the signal phasing that is in operation and the traffic conditions at different times throughout the day. It is known in Perth that the public express that they like to have longer green times rather than short phases but shorter phases means that on average people will experience less delay and thus communication around any change to cycle times will be key.

The maximum cycle times on the Orrong Road corridor are again dependent on the time of day but in the AM and PM periods a maximum cycle time up to 155 seconds is advised.

Incremental Split Selection Secondly, switching to incremental split selection, a function within SCATS, allows the signals to be more adaptive to the detected traffic volumes which generally improves the signal performance compared to using fixed signal plans. For this reason average phase splits have not been provided from the modelling analysis. From the Orrong Road Option 5 analysis, incremental split selection reduced delay, particularly in the peak periods.

Master-isolated operation Specifically relating to the Tonkin Highway/ Kelvin Road intersection, a change from Flexi-link to Master-isolated (both of which are SCATS operation settings) shows improvements, particularly if it is carried out in conjunction with double diamond overlap phasing. This does have an impact on the Tonkin Highway north approach right turn movement which is currently a safety issue on-street and the length of this turn bay is a factor in the success of this SCATS setting at this demonstration site.

SCATS signal settings guidance There is little useful guidance or tools within the industry that communicate the range of SCATS settings that may be employed by operators, and the time periods for which they are appropriate. As a result, when traffic engineers at the TOC make changes to the operation of a set of signals they use

Project 232984 9 October 2013 Revision 2 Page 35

techniques developed from their signals experience and these techniques include reviewing traffic patterns at the signalised intersection as each site has its own unique characteristics and behaviours.

8.1.2 Main Roads WA TOC procedure

Signal auditing As is common in all TOC’s, operators observe the network to identify sites where congestion can be reduced by balancing the delay on intersection approaches. This leads to adjustment of the SCATS signal settings to the point where the on-street performance is deemed to be optimal.

It is evident from the workshops with Main Roads WA that the TOC is very responsive to comments from both the public and from other road authorities. To some extent complaints made to Main Roads WA drive which signalised intersections or corridors will be reviewed. This aligns with Main Roads WA philosophy of valuing and prioritising the experience of road network users.

A secondary driver for undertaking signal reviews is changes in land use, particularly new developments that result in changes to the road network. These will generally change traffic patterns and trigger a need to audit nearby SCATS signals that may have been impacted by the change. Audits relating to these types of changes are generally planned in advance.

This approach is however reactive and the more proactive, and more resource intensive, approach of auditing of signals currently occurs approximately once every three years.

Figure 8-1 shows the process that the TOC currently apply to audit and adjust SCATS signal settings.

Figure 8-1: Main Roads WA signal adjustment process

1. Indication that a change to signal settings is needed

Indication provided by either a member of the public, another government authority, or by a TOC operator

2. Site visit

Undertaken as an audit during the peak periods by TOC staff once every two to three years for each site

3. SCATS data

Regularly collected and analysed by TOC staff

4. SIDRA analysis

If the signal intersection design change is considered major, SIDRA analysis is required before on street implementation

5. Signal settings adjustment

Signal settings adjusted until a balance of the delay across all of the approaches is achieved

Project 232984 9 October 2013 Revision 2 Page 36

Figure 8-2 shows the process applied to audit and adjust SCATS signal settings, as used in this study. The primary differences between the two methodologies are the additional steps of collecting on-site traffic count data and the development of an operational micro-simulation model to look at both individual intersections and corridors.

Figure 8-2: Signal adjustment process used for this study

1. Sites selected by relevant organisation/authority

2. Site visit undertaken during the peak periods

3. SCATS data collected and analysed

4. Traffic counts undertaken at the site

5. Base traffic model built

6. Integration of SCATS signal control

7. Calibration of the models to existing on-street conditions

8. Development of major and minor SCATS and intersection design options for each site

9. Options tested / optimised

10. Optimal signal settings then implemented on-street and their performance assessed

Austroads Guide to Traffic Management Part 9 identifies five different audit levels with Level 3, Strategic Data Audit, and Level 4, Minor Network Control, most applicable to this study. They generally advise site visits and analysis of scats data to identify which of the wide range of SCATS settings should be adjusted.

Project 232984 9 October 2013 Revision 2 Page 37

8.1.3 Testing of possible refinement options before implementation The implementation of a wider range of adjustments such as the new phasing arrangements along the Orrong Road corridor, not only has the potential to improve signal operation but equally the potential to have a detrimental impact on the road network. Additionally, because permanent civil works are sometimes required, confidence is needed that the adjustments will result in improvements.

This is where the benefit of using intersection and network analysis software programs to predict the impact of signal and design changes is demonstrated. For most new installations, Main Roads WA currently uses SIDRA to assist in the selection of the phasing. However, this uses fixed formulas for the analysis rather than an operational model such as micro-simulation, which replicates observed on- street driver behaviour.

A further benefit of an operational model is that quantitative analysis can be carried out later, beyond the time when signals are installed, to take account of continuing changes in land use and traffic flow. Micro-simulation, as used in this study, is the next step forward as it allows SCATS signal control to be used within the virtual environment and it models individual vehicles in an operational format to provide a high level of confidence about what the outcomes will be. However, micro-simulation models take more time to build, require more data, and often require modelling personnel.

As shown in this study, analysis of some options may show that the existing on-street signal setup operates better than some alternate options that are proposed and that the infrastructure is performing close to its optimal performance level.

Conversely analysis showed that positive improvements can potentially be achieved without civil works. Regardless of how slight some of these improvements may be they are nonetheless justified by extending the life of the existing infrastructure and thus the additional investment in micro-simulation analysis is recouped.

There is an opportunity for Main Roads WA to use analysis techniques to assist in getting the most out of the existing infrastructure. Many road agencies around the world invest in modelling or analysis teams to work alongside the TOC to ensure the optimal signal setup is used not only at the time of signal installation, but also at regular intervals during the signals lifetime.

Real-time incident modelling The next generation of using modelling to simulate real life network operation is real-time incident modelling. This works by inputting the incident as a road network ‘blockage’ into a prebuilt model. The model then quickly runs several network scenarios producing outputs which advise TOC operator’s how best to manage traffic so the impact of the incident on the network can be minimised.

8.1.4 Resourcing

TOC staff numbers The Main Roads WA TOC currently has 6 staff each covering approximately 150 intersections. More staff with high skill sets would allow signals and corridors to be audited in accordance with Austroads guidelines at more frequent intervals. This would ultimately result in a network that is operating closer to its optimum level. In the NSW RMS, TOC operators cover between 250 and 300 sites each, however these sites are located in a more condensed pattern than in Perth, making it easier for staff to reach sites for on-street monitoring and auditing.

Project 232984 9 October 2013 Revision 2 Page 38

The ‘Traffic Signal Operations and Maintenance Staffing Guidelines’, made available by the US Department of Transportation Federal Highway Administration (FHWA), states that 75 to 100 signals per operator/ traffic engineer are recommended to achieve desired efficiency targets. However the use of fixed time signal is more prevalent in the US, as opposed to the SCATS adaptive system, and these fixed time signals generally require more operator involvement thus more operators are required.

Industry constraints Attracting and retaining TOC staff with the appropriate skill set is no easy task. In today’s environment, SCATS operators must not only be highly competent in SCATS signal setup, monitoring, and maintenance but must also have a good understanding of traffic engineering and the related concepts. This is to ensure that staff take into consideration the impact of their signal decision on the rest of the road network as well as other road users.

Traditionally many SCATS operators have come from an electrical engineering background and have been trained ‘on the job’. Today there are various courses available for new staff to gain knowledge in this field. However with SCATS operation being so specialised, it is believed that some staff fear being pigeon-holed as a SCATS specialist.

The FWHA guidelines state that, as a minimum, new technicians should have High School Qualifications, knowledge of electrical aspects, and initial certification from a traffic signals course. At the other end of the scale, the traffic system engineering team leader should be professional registered as an engineer, preferably with a traffic operations engineer certification of some sort.

Project 232984 9 October 2013 Revision 2 Page 39

9. Recommendations

The study examined two demonstration sites and Main Roads WA signal processes to determine the following:

 What signalised intersection treatments are currently utilised.  What signal improvements can be achieved.  General considerations and requirements for refining SCATS signal settings.  Signal refinement methodologies.  Performance measures and evaluation of signal timing performance.  Barriers to refining signal settings.  Main Roads WA signal refinement policy, management, planning, and resources. From this study the following items are recommended for implementation at demonstration sites and for further investigation for the wider Main Roads WA network. These items would allow the SCATS signal system to be used to its full potential.

1. To reduce delay, lower cycle times should be implemented at the demonstration sites. No single ideal cycle time number exists as it depends on the signal phasing at the site and the traffic conditions which change throughout the day. Research to further develop an understanding of the driver behaviour that leads to this relationship would be useful. 2. The use of incremental split selection is recommended. Incremental split selection is a SCATS setting that allows a more dynamic selection of split plans and thus more dynamic signal operation to further assist the balancing approaches and reducing delay. 3. A change to master-isolated settings should be investigated at Flexilink sites. Master-isolated signals allow maximum green times for each phase to be determined by SCATS based on the approach degree of saturation. Flexi-link operates differently using predefined plans and time of day schedules. 4. Some of the signal refinement options identified as having efficiency benefits should be trialled on- street. Data would need to be collected before and during the trials across a number of days during the critical time periods with final adjustments also made once in operation on-street. Post- implementation data such as travel time data can be collected once the new signal settings are fully operational. The SCATS settings adjusted in the Orrong Road corridor Option 5 test, particularly the use of incremental split selection, would be the best to trial first on-street due to no civil works required. 5. More detailed operational analysis using micro-simulation modelling within the TOC methodology would allow for a more diverse range of options to be tested in order to achieve the best possible level of performance from the existing road network. Improvements can be tested for the most conventional of signalised intersections to the very complex. 6. Since resource constraints are the most significant factor contributing to sub-optimal signal retiming it is recommended that there is a reduction in the average number of signals monitored per TOC operator. 7. Investigation into new signal refinement approaches and tools such as real-time incident modelling should be continued. 8. It is advised that if one approach works at a selected intersection, it should not be automatically applied to others. Staff at the TOC should follow a robust methodology to assess what changes may be required on a site by site basis, as each site has their own characteristics and reacts accordingly. This important point was displayed in this study with some of the tested options worsening the performance of the respective demonstration sites.

Project 232984 9 October 2013 Revision 2 Page 40

9. There is little useful national guidance or tools that communicate the range of SCATS settings that may be employed by operators, and the time periods for which they are appropriate. As a result, traffic engineers at the TOC use techniques based on experience and by reviewing traffic patterns as each site has its own unique characteristics and behaviours. Research is also needed to develop guidance based on intersection performance measures. 10. Data is a crucial factor in being able to assess if signals are currently performing satisfactorily and if refinements are having the desired outcomes. To provide a basis for improved signal refinement, collection and calculation of the following data is recommended: - Total volume throughput - Degree of saturation - Average corridor speeds - Spillback of queues through turn bays or to nearby intersections - Delay caused to other movements by these queues spilling back 11. The Main Roads WA TOC delivers a wide range of network management services, and obtaining and retaining sufficient numbers of skilled staff within budget constraints is a difficult task. It is recommended that Main Roads WA compare their signal operations with NSW RMS and other Australian States. This will assist Main Roads WA in ensuring that their team of staff maintain the highest skill set possible. This is particularly relevant with the wide range of new Intelligent Transport Systems technologies that are soon to be trialled in Perth which will further stretch Main Roads WA TOC resources.

Project 232984 9 October 2013 Revision 2 Page 41

Appendix A Further Crash Data

A. Tonkin Highway / Kelvin Road

A total of 203 crashes, involving all modes of transport, occurred over the 2008 – 2012 period at the Kelvin Road - Tonkin Highway intersection.

A.1 Crash Severity The number of crashes and severity per year is shown in Table 4-1. In summary:

 A total of 203 crashes occurred on the intersection  4.5% resulted in hospital treatment  16% resulted in medical treatment  53% resulted in major property damage only (PDO Major)  26.5% resulted in minor property damage only (PDO Minor)

It is observed that more than 20% of the crashes resulted in either medical or hospital treatment.

A.2 Time of Crash For detailed time analysis of the data, only the Monday to Friday data was used, however, Saturday and Sunday statistics are also included. Nearly 50% of the total number of crashes occurred during the morning (6am – 10am) and afternoon (2pm – 5pm) peaks. This suggests that congestion of the intersection plays an important factor in the time during which a crash occurred.

Detailed summary of the time of crash occurrences

Time of Crash Number of Crashes Number of Crashes as a percentage

Morning 6 am – 10 am 57 28%

Midday 10 am – 2 pm 26 13%

Afternoon 2 pm – 5pm 42 21%

Early Evening 5 pm – 7 pm 22 11%

Other 7 pm – 6 am 21 10%

Saturday and Sunday 35 17%

A.3 Vehicle Type and Crash Movement Summary The tables below summaries the type of vehicle involved in the crash and the crash movement. Nearly 90% of vehicles involved in a crash were those that involved a car. The majority of the crashes occurred when the vehicle was either fully stopped or travelling straight (not out of control).

Type of vehicle involved in crashes

Vehicle Type Number of Crashes Number of Crashes as a total Percentage

Motorcycle 5 2.5%

Cars 178 88%

Bus 1 0.5%

Truck 6 3%

Unclassified 13 6%

Crash movement summary

Description of Crash Movement Number of Crashes at Intersection

Out of control 7

Overtaking 7

Reversing or Rolling Back 1

Stopped 72

Turning 8

Swing Wide 1

Straight ahead – not out of control 107

TOTAL 203

A.4 Nature of Crash Event The crash statistics data represented in the table below shows the nature of the event for the particular vehicle movement. From this we identify the Tonkin Highway North – South and South – North movements as the prevailing crash movements with the highest proportion of these being rear endings.

Nature of crash at Tonkin Highway / Kelvin Road

Nature of Event

Movement Rear End Right Angle Right Turn Sideswipe Undefined TOTAL Through Same Direction

E-S 2 2

E-W 1 1

N-W 7 4 1 12

N-S 21 2 2 3 1 29

N-E 1 1

N-undefined 1 1

W-N 5 5

W-E 2 2 4

S-E 9 1 10

S-N 22 1 23

S-W 3 3

S-undefined 1 1

Undefined- 6 1 1 8 Undefined

TOTAL 80 6 6 6 2 100

B. Orrong Road

B.1 Crash Severity The number of crashes and severity per year is shown in Table 4-3. In summary:

 A total of 1493 crashes occurred along the corridor  0.15% resulted in a fatality  4% resulted in hospital treatment  15.85% resulted in medical treatment  50% resulted in major property damage only (PDO Major)  30% resulted in minor property damage only (PDO Minor)

It is observed that 20% of the crashes resulted in either medical or hospital treatment. Alarmingly, a clear upward trend, in terms of the number of crashes, can be identified from the data. The reason for the trend is not clear but an increase in the number of vehicles on our roads is a likely contributing factor. More than 50% of the crashes resulted in major property damage only.

B.2 Time of Crash For detailed time analysis of the data, only the Monday to Friday data was used, however, Saturday and Sunday statistics are also included. 46% of the total number of crashes occurred during the morning and afternoon peaks. This suggests that congestion along the corridor plays an important factor in the time during which a crash occurred.

Detailed summary of the time of crash occurrence

Time of Crash Number of Crashes Number of Crashes as a percentage

Morning 6 am – 10 am 331 22.17 %

Midday 10 am – 2 pm 193 12.93 %

Afternoon 2 pm – 5pm 356 23.84 %

Early Evening 5 pm – 7 pm 211 14.13 %

Other 7 pm – 6 am 134 8.97 %

Saturday and Sunday 268 17.96 %

B.3 Vehicle Type and Crash Movement Summary The tables below summaries the type of vehicle involved in the crash and the crash movement. Nearly 90% of vehicles involved in a crash were those that involved a car. For both crashes that occurred at an intersection and for those that occurred midblock of two intersections, the majority of the crashes happened when the vehicle was either fully stopped or travelling straight (not out of control).

Type of vehicle involved in the crash

Vehicle Type Number of Crashes Number of Crashes as a total Percentage

Pedestrians 2 0.13 %

Bicycle 8 0.54 %

Motorcycle 19 1.27 %

Cars 1314 88.01 %

Bus 7 0.47 %

Truck 38 2.55 %

Unclassified 105 7.03 %

Crash movement summary

Description of Crash Movement Number of Crashes at Number of Crashes Intersection Midblock

Out of control 19 10

Overtaking 18 22

Reversing or Rolling Back 12 1

Stopped 346 86

Turning 165 14

Swing Wide 5 1

Swerving 3 2

Straight ahead – not out of control 575 214

TOTAL 1143 350

B.4 Nature of Crash Event This is the data collected for the direction of the target vehicle only. The colliding vehicle direction was ignored within this summary. Note that Rivervale Wattle Grove Link is referred to as Orrong Road on the surveyed data.

B.4.1 Francisco Street / Francisco Place / Orrong Road The table below represents the nature of the crash event and the respective given movement at the Francisco Street, Francisco Place and Orrong Road intersection. From this we identify rear end crashes as the dominant type of crash with the West – East and East – West movements as the prevailing crash movements.

Nature of crash at Francisco Street / Francisco Place / Orrong Road

Nature of Event

Movement Rear End Right Angle Right Turn Sideswipe Undefined TOTAL Through Same Direction

W-E 33 2 1 36

W-N 2 2

W-S 6 6

E-W 25 1 2 28

E-N 2 11 1 14

N-undefined 2 2

N-W 2 2

N-E 3 1 4

Undefined 3 2 5

TOTAL 72 3 21 1 2 99

B.4.2 Archer Street / Orrong Road The table below represents the nature of crashes and the respective given movement at Archer Street and Orrong Road. The most number of crashes occurred on the East – West direction, where the majority of crashes were rear end. Right turn through crashes was dominant in the West – South direction.

Nature of crash at Archer Street / Orrong Road

Nature of Event

Movement Rear End Right Angle Right Turn Sideswipe Undefined TOTAL Through Same Direction

W-E 8 8

W-S 1 18 2 19

E-W 28 1 9 1 39

E-S 6 6

S-E 2 3 5

S-W 4 1 1 6

S-undefined 2 1 3

Undefined 4 4

TOTAL 54 7 27 1 3 92

B.4.3 Wright Street / Orrong Road The table below represents the nature of crashes and the respective given movement at Wright Street and Orrong Road. Rear end crashes were the most dominant in both the West – East and East – West directions.

Nature of crashes at Wright Street / Orrong Road

Nature of Event

Movement Rear End Right Angle Right Turn Sideswipe Undefined TOTAL Through Same Direction

W-E 21 1 1 23

W-N 1 1 2

W-undefined 1 1

E-W 19 1 20

E-N 4 1 5

N-E 1 1 2

N-W 3 2 1 1 7

N-undefined 1 1 2 4

Undefined 1 1

TOTAL 52 4 2 4 3 65

B.4.4 Oats Street / Orrong Road The table below represents the nature of crashes and the respective given movement at Oats Street and Orrong Road. Rear end crashes were the most dominant in both the West – East and East – West directions.

Nature of crashes at Oats Street / Orrong Road

Nature of Event

Movement Rear End Right Angle Right Turn Sideswipe Undefined TOTAL Through Same Direction

W-E 40 3 43

W-S 2 1 3

E-W 51 1 52

E-S 1 1

E-N 1 1 2

E-undefined 2 2

N-S 3 1 1 5

N-W 5 1 6

N-E 1 1

S-N 4 1 3 2 10

S-W 1 1

S-E 1 1

S-undefined 2 2

Undefined 4 2 6

TOTAL 113 2 10 10 135

Appendix B Calibration Details

y = 0.9417x + 15.558 R² = 0.9967 Tonkin Highway/ Kelvin Road Turn Count Comparison - AM Peak Hour

2500

2000

1500

Modelled 1000

500

0 0 200 400 600 800 1000 1200 1400 1600 1800

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 1.0298x - 3.5737 R² = 0.9998 Tonkin Highway/ Kelvin Road Turn Count Comparison - Midday Peak Hour

1200

1000

800

600 Modelled

400

200

0 0 100 200 300 400 500 600 700 800 900

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 1.0085x - 1.7854 R² = 0.999 Tonkin Highway/ Kelvin Road Turn Count Comparison - PM Peak Hour

3000

2500

2000

1500 Modelled

1000

500

0 0 500 1000 1500 2000 2500

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 1.0231x - 0.608 R² = 0.9997 Tonkin Highway/ Kelvin Road Turn Count Comparison - Offpeak

600

500

400

300 Modelled

200

100

0 0 50 100 150 200 250 300 350 400 450

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 1.0106x + 2.1765 R² = 0.9901 Orrong Road Corridor Turn Count Comparison - AM Peak Hour

3500

3000

2500

2000

Modelled 1500

1000

500

0 0 500 1000 1500 2000 2500

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 0.995x + 0.5411 R² = 0.9966 Orrong Road Corridor Turn Count Comparison - Midday Peak Hour

2500

2000

1500

Modelled 1000

500

0 0 200 400 600 800 1000 1200 1400 1600 1800

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 0.9881x - 2.1407 R² = 0.9964 Orrong Road Corridor Turn Count Comparison - PM Peak Hour

3000

2500

2000

1500 Modelled

1000

500

0 0 500 1000 1500 2000 2500

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

y = 1.0038x + 2.3464 R² = 0.9937 Orrong Road Corridor Turn Count Comparison - Offpeak

1200

1000

800

600 Modelled

400

200

0 0 100 200 300 400 500 600 700 800 900

Observed

Observed vs Modelled x = y +30% -30% Linear (Observed vs Modelled)

Appendix C Modelling Delay Outputs

Level of Service Legend (seconds/vehicle)

LOS From To A 0 10 B 10 20 C 20 35 D 35 55 E 55 80 F 80 <

Tonkin Highway/ Kelvin Road Base Model Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 22 21 22 5 12 13 69 8 Tonkin N T 849 856 1956 413 17 15 80 8 R 447 318 310 92 59 28 96 27 L 81 40 89 23 87 18 72 10 Kelvin E T 170 62 91 33 157 74 104 41 R 25 22 14 1 160 79 119 37 Tonkin/ Kelvin L 189 96 97 15 197 7 27 0 Tonkin S T 1546 767 1001 292 221 49 67 13 R 31 27 87 19 311 87 152 44 L 412 322 536 82 45 21 70 19 Kelvin W T 85 58 197 32 106 70 118 34 R 75 88 288 24 105 67 116 32

Tonkin Highway/ Kelvin Road Option 1 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 22 21 21 5 15 14 125 8 Tonkin N T 849 854 1803 413 17 14 140 7 R 449 317 286 93 84 27 144 20 L 83 40 89 23 44 15 38 10 Kelvin E T 176 68 98 32 96 50 65 45 R 23 22 14 1 106 50 66 37 Tonkin/ Kelvin L 183 96 97 15 191 4 8 1 Tonkin S T 1514 774 1007 293 218 35 39 14 R 30 26 90 19 249 54 66 42 L 410 320 542 81 38 17 32 16 Kelvin W T 86 59 205 32 84 43 68 39 R 75 89 296 24 79 42 67 35

Tonkin Highway/ Kelvin Road Option 2 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 22 21 22 5 12 14 37 8 Tonkin N T 845 860 1958 414 13 14 45 5 R 447 318 309 94 49 23 50 23 L 69 39 94 23 416 20 88 10 Kelvin E T 141 71 115 34 460 56 114 29 R 22 22 15 1 404 72 110 37 Tonkin/ Kelvin L 194 96 97 15 135 5 14 0 Tonkin S T 1573 775 994 294 159 40 50 10 R 32 27 88 19 217 69 88 35 L 413 320 482 81 37 16 106 13 Kelvin W T 87 59 182 32 83 47 162 24 R 75 87 219 24 95 58 422 35

Tonkin Highway/ Kelvin Road Option 3 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 22 21 22 5 12 13 71 8 Tonkin N T 849 857 1919 413 16 15 85 8 R 444 317 306 92 43 23 88 27 L 82 40 90 23 59 15 54 10 Kelvin E T 172 72 109 34 125 67 93 41 R 24 22 14 1 124 67 88 37 Tonkin/ Kelvin L 198 96 96 15 31 3 8 0 Tonkin S T 1630 779 993 293 65 43 58 13 R 32 26 90 19 150 81 102 43 L 408 319 547 82 45 19 33 20 Kelvin W T 83 58 205 32 95 56 82 32 R 74 87 295 24 92 55 79 32

Tonkin Highway/ Kelvin Road Option 4 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 22 21 21 5 21 13 101 8 Tonkin N T 838 858 1868 413 26 12 113 5 R 443 317 299 94 118 36 132 23 L 80 40 90 23 103 15 36 10 Kelvin E T 169 71 110 34 143 35 51 31 R 22 22 14 1 131 50 70 37 Tonkin/ Kelvin L 189 96 97 15 92 4 6 0 Tonkin S T 1533 778 998 294 115 28 28 9 R 31 26 90 19 133 50 50 32 L 412 321 546 82 31 15 28 14 Kelvin W T 89 59 207 32 54 32 50 25 R 74 87 296 24 78 41 1 34

Orrong Road Corridor Base Model Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 60 134 134 134 18 22 2 1 Francisco N T 2 0 1 1 224 0 0 0 R 473 359 333 93 214 57 120 21 L 2 7 7 3 18 22 2 1 Orrong E T 2086 1521 1893 631 22 21 10 6 R 154 158 167 67 49 29 24 5 Francisco/Orrong L 6 11 6 2 86 72 81 36 Francisco S T 5 3 3 2 72 56 87 25 R 7 12 6 2 58 66 84 22 L 325 288 316 142 35 15 68 1 Orrong W T 1837 1505 1941 870 48 31 80 7 R 7 17 10 1 53 26 68 2 L 95 120 133 17 61 30 41 8 Orrong E T 1790 1438 1757 498 62 30 40 6 L 426 184 298 132 65 55 54 16 Archer/Orrong Archer S R 229 204 210 61 76 46 69 26 T 1534 1251 1503 676 16 9 5 4 Orrong W R 211 172 172 103 56 38 56 4 L 119 119 121 37 56 38 82 22 Wright N R 283 237 260 77 73 61 124 34 T 1646 1318 1620 438 58 2 22 3 Wright/Orrong Orrong E R 115 134 155 51 101 52 75 38 L 169 194 234 67 14 3 18 2 Orrong W T 1648 1314 1597 685 30 12 24 8 L 84 89 123 29 227 43 77 27 Oats N T 400 162 305 37 230 57 91 32 R 63 54 37 20 222 57 89 33 L 34 38 43 21 151 30 82 16 Orrong E T 1622 1297 1675 429 145 28 80 13 R 87 118 155 36 194 72 149 42 Oats/Orrong L 97 92 89 36 66 33 153 18 Oats S T 238 178 371 81 81 43 159 20 R 148 75 142 48 123 52 196 24 L 30 29 32 30 80 42 34 18 Orrong W T 1608 1284 1573 653 79 43 31 17 R 142 119 114 36 136 73 89 42

Orrong Road Corridor Option 1 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 66 134 134 134 6 8 13 0 Francisco N T 2 0 1 1 98 0 0 0 R 508 359 336 93 98 49 61 30 L 3 9 7 3 6 8 13 0 Orrong E T 2039 1498 1863 632 19 13 10 4 R 144 158 169 67 35 22 34 4 Francisco/Orrong L 6 11 6 2 78 66 80 49 Francisco S T 5 3 3 2 96 61 79 51 R 7 12 6 2 81 53 62 34 L 323 288 316 142 87 13 42 1 Orrong W T 1827 1505 1939 870 104 29 55 6 R 7 17 10 1 106 27 42 3 L 97 119 132 17 33 49 22 4 Orrong E T 1811 1436 1738 499 35 52 19 3 L 426 184 298 132 73 71 56 33 Archer/Orrong Archer S R 229 204 210 61 84 39 66 41 T 1503 1256 1503 674 5 35 6 4 Orrong W R 213 171 174 103 74 68 46 3 L 119 119 121 37 53 40 40 21 Wright N R 283 237 260 77 87 92 68 37 T 1675 1314 1614 438 26 11 24 2 Wright/Orrong Orrong E R 114 134 153 50 80 44 67 36 L 165 195 233 67 17 44 21 3 Orrong W T 1660 1304 1609 686 21 54 24 5 L 83 89 123 29 250 36 55 30 Oats N T 398 162 305 37 256 50 67 37 R 64 54 37 20 252 51 62 38 L 34 37 44 22 73 32 212 18 Orrong E T 1641 1303 1646 429 75 27 204 13 R 88 121 152 36 124 59 258 44 Oats/Orrong L 97 92 91 36 62 31 63 22 Oats S T 239 178 375 81 77 38 70 23 R 149 75 146 48 123 45 102 21 L 28 29 32 30 54 25 29 15 Orrong W T 1600 1263 1569 652 53 25 24 13 R 144 119 115 36 144 61 74 50

Orrong Road Corridor Option 2 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 60 134 134 134 23 20 2 0 Francisco N T 2 0 1 1 264 0 0 0 R 468 359 332 93 247 59 120 20 L 2 8 6 3 23 20 2 0 Orrong E T 2079 1520 1872 630 19 20 9 7 R 152 157 167 67 50 30 26 5 Francisco/Orrong L 6 11 6 2 71 58 85 33 Francisco S T 5 3 3 2 89 67 99 32 R 7 12 6 2 74 66 74 27 L 325 288 317 142 38 16 46 1 Orrong W T 1837 1505 1941 870 49 33 59 7 R 7 17 10 1 46 30 44 2 L 96 120 135 17 62 28 39 6 Orrong E T 1764 1439 1731 496 63 29 37 7 L 426 184 298 132 60 55 60 16 Archer/Orrong Archer S R 229 204 210 61 73 48 70 27 T 1538 1250 1505 675 16 9 5 4 Orrong W R 210 171 174 104 59 38 51 4 L 119 119 120 37 62 38 110 24 Wright N R 283 237 258 77 91 61 157 34 T 1641 1314 1605 439 76 2 21 3 Wright/Orrong Orrong E R 111 135 155 51 111 52 77 37 L 167 193 235 67 15 4 17 3 Orrong W T 1654 1313 1594 684 30 12 22 7 L 77 89 123 29 297 47 116 28 Oats N T 376 162 305 37 294 59 131 36 R 58 54 37 20 294 57 119 30 L 34 37 43 22 94 28 62 16 Orrong E T 1637 1301 1680 429 94 27 63 14 R 88 121 155 36 140 70 132 42 Oats/Orrong L 96 92 89 36 83 36 155 19 Oats S T 237 178 366 81 103 48 163 23 R 148 75 141 48 137 50 196 23 L 30 30 33 30 67 43 29 18 Orrong W T 1610 1284 1570 653 67 42 27 17 R 144 118 116 35 119 79 91 46

Orrong Road Corridor Option 3 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 66 134 134 134 19 4 14 0 Francisco N T 2 0 3 1 102 0 0 0 R 508 359 195 93 111 47 73 32 L 2 9 5 3 19 4 14 0 Orrong E T 2002 1503 1871 631 49 14 13 4 R 138 157 163 67 63 17 28 5 Francisco/Orrong L 17 25 15 6 75 57 77 50 Francisco S T 0 0 0 0 0 0 0 0 R 0 0 0 0 0 0 0 0 L 325 288 316 142 36 10 45 1 Orrong W T 1837 1505 1942 870 50 23 59 6 R 7 17 10 1 50 18 50 0 L 96 119 134 17 29 50 58 3 Orrong E T 1787 1434 1755 499 29 50 57 3 L 411 184 298 132 135 74 53 33 Archer/Orrong Archer S R 221 204 210 61 124 36 74 40 T 1535 1254 1536 675 10 20 9 4 Orrong W R 215 171 259 103 68 53 72 4 L 117 119 121 37 107 48 84 24 Wright N R 280 237 260 77 110 68 86 42 T 1633 1322 1645 437 24 7 57 2 Wright/Orrong Orrong E R 114 135 157 51 85 47 99 40 L 162 195 236 67 3 8 12 1 Orrong W T 1625 1315 1595 683 8 15 16 3 L 83 89 123 29 252 40 85 30 Oats N T 393 162 305 37 258 50 96 40 R 62 54 37 20 255 52 93 32 L 34 38 44 21 99 29 105 14 Orrong E T 1604 1303 1686 430 101 28 104 13 R 86 120 154 36 147 57 190 44 Oats/Orrong L 97 92 88 36 69 27 172 22 Oats S T 239 178 367 81 82 37 180 23 R 149 75 142 48 130 41 226 25 L 29 30 32 30 62 32 24 13 Orrong W T 1567 1283 1563 651 61 32 22 12 R 141 119 114 36 143 66 84 49

Orrong Road Corridor Option 4 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 60 134 134 134 14 19 3 1 Francisco N T 2 0 1 1 254 0 0 0 R 469 359 333 93 243 57 109 21 L 2 8 7 3 14 19 3 1 Orrong E T 2070 1510 1876 631 20 21 9 7 R 156 157 168 67 51 28 26 6 Francisco/Orrong L 6 11 6 2 76 69 86 26 Francisco S T 5 3 3 2 92 57 80 46 R 7 12 6 2 68 58 77 32 L 325 288 316 142 32 17 58 1 Orrong W T 1836 1505 1943 870 43 33 71 7 R 7 17 10 1 39 29 64 3 L 95 119 134 17 54 36 41 6 Orrong E T 1758 1431 1724 497 55 36 40 7 L 426 184 298 132 63 56 57 16 Archer/Orrong Archer S R 229 204 210 61 81 47 71 28 T 1537 1255 1506 676 24 9 5 4 Orrong W R 210 172 174 103 60 38 50 5 L 119 119 121 37 68 37 80 21 Wright N R 283 237 260 77 92 59 118 33 T 1602 1325 1599 438 20 2 24 3 Wright/Orrong Orrong E R 112 132 149 51 70 51 78 37 L 167 194 233 67 24 3 19 2 Orrong W T 1650 1316 1609 686 43 11 24 7 L 73 88 116 29 287 49 208 25 Oats N T 351 160 288 37 287 60 215 36 R 57 54 37 20 266 41 190 20 L 32 38 43 21 249 35 183 12 Orrong E T 1584 1297 1665 431 246 34 180 12 R 85 120 151 36 295 77 251 41 Oats/Orrong L 96 90 81 36 78 52 218 26 Oats S T 234 176 330 81 97 66 227 31 R 149 75 136 48 124 53 232 23 L 29 29 31 30 107 50 60 14 Orrong W T 1569 1278 1572 654 108 48 59 13 R 142 120 115 35 158 75 104 47

Orrong Road Corridor Option 5 Delay (seconds/vehicle)

Name Approach Movement AM Volume MD Volume PM Volume OP Volume AM Delay MD Delay PM Delay OP Delay L 66 134 134 134 10 1 1 1 Francisco N T 2 0 1 1 105 0 0 0 R 508 359 336 93 90 48 67 29 L 3 9 8 3 10 1 1 1 Orrong E T 2052 1485 1898 630 17 13 12 4 R 142 156 169 67 43 21 35 5 Francisco/Orrong L 6 11 6 2 71 56 67 47 Francisco S T 5 3 3 2 67 62 66 38 R 7 12 6 2 74 54 68 34 L 325 288 316 142 77 15 44 1 Orrong W T 1836 1505 1937 870 93 31 56 7 R 7 17 10 1 102 29 43 4 L 99 117 136 17 37 61 50 3 Orrong E T 1834 1411 1763 497 38 60 50 3 L 426 184 298 132 81 61 57 32 Archer/Orrong Archer S R 229 204 210 61 83 36 66 42 T 1517 1255 1502 674 10 23 5 4 Orrong W R 213 171 174 103 110 64 70 4 L 119 119 121 37 57 38 57 27 Wright N R 283 237 260 77 89 77 82 38 T 1682 1318 1638 438 25 18 29 1 Wright/Orrong Orrong E R 115 134 155 51 83 59 86 40 L 166 196 236 67 20 17 8 2 Orrong W T 1665 1318 1614 686 22 23 11 4 L 83 89 123 29 250 38 78 30 Oats N T 396 162 305 37 255 53 94 37 R 63 54 37 20 255 54 87 36 L 34 38 44 22 80 32 53 17 Orrong E T 1642 1301 1682 431 78 29 52 13 R 88 120 154 36 123 65 151 43 Oats/Orrong L 97 92 88 36 68 28 162 20 Oats S T 240 178 366 81 79 37 166 24 R 149 75 141 48 126 43 205 24 L 29 29 32 30 56 32 27 11 Orrong W T 1598 1273 1587 654 57 33 29 12 R 144 119 114 36 147 68 100 44

Appendix D SCATS Counts Comparison

The Austraffic traffic counts and SCATS detector counts were compared for both the Tonkin Highway / Kelvin Road intersection, and the Orrong Road corridor and are shown below for the AM peak, Midday peak, PM peak and Off peak periods. These results show that the SCATS counts are within 10% of traffic surveys and can then be considered relatively accurate. However SCATS counts do not capture individual movements in shared lanes.

Traffic count and signal data comparison for Tonkin Highway / Kelvin Road intersection AM PEAK 6:45 - 7:45 AM Austraffic Intersection Approach SCATS count % Difference Count North 1285 1245 3.11% West 565 530 6.19% Tonkin Highway / Kelvin Road South 1843 1665 9.66% East* 215 188 12.56% Total 3908 3628 7.16% MIDDAY PEAK 11.45 AM - 12:45 PM Austraffic Intersection Approach SCATS count % Difference Count North 1163 1129 2.92% West 465 423 9.03% Tonkin Highway / Kelvin Road South 878 769 12.41% East* 98 94 4.08% Total 2604 2415 7.26% PM PEAK 4:00 - 5:00 PM Austraffic Intersection Approach SCATS count % Difference Count North 2268 2225 1.90% West 1063 1007 5.27% Tonkin Highway / Kelvin Road South 1129 1037 8.15% East* 117 111 5.13% Total 4577 4380 4.30% OFFPEAK 9:00 - 10:00 PM Austraffic Intersection Approach SCATS count % Difference Count North 502 485 3.39% West 135 130 3.70% Tonkin Highway / Kelvin Road South 322 308 4.35% East* 30 28 6.67% Total 989 951 3.84% *Note that this approach does not include the left turn movement out of Kelvin Road

Traffic count and signal data comparison for the Orrong Road corridor AM PEAK 7:15 - 8:15 AM Austraffic Intersection Approach SCATS count % Difference Count West 1692 1660 1.89% South 506 581 -14.82% Orrong Road / Oats Street East 1535 1499 2.35% North 559 522 6.62% West 1792 1809 -0.95% Orrong Road / Wright Street East 1595 1550 2.82% North 423 414 2.13% West 1926 1838 4.57% Orrong Road / Archer Street South 670 674 -0.60% East 1805 1805 0.00% West 2145 1822 15.06% South 18 14 22.22% Orrong Road / Francisco Street East 2410 2345 2.70% North 568 - N/A Total 17076 16533 3.18% MIDDAY PEAK 11.30 AM - 12:30 PM Austraffic Intersection Approach SCATS count % Difference Count West 1408 1420 -0.85% South 350 426 -21.71% Orrong Road / Oats Street East 1447 1428 1.31% North 288 288 0.00% West 1588 1559 1.83% Orrong Road / Wright Street East 1469 1465 0.27% North 371 346 6.74% West 1590 1587 0.19% Orrong Road / Archer Street South 392 375 4.34% East 1579 1560 1.20% West 1796 1489 17.09% South 25 21 16.00% Orrong Road / Francisco Street East 1707 1673 1.99% North 399 - N/A Total 14010 13637 2.66%

PM PEAK 4:15 - 5:15 PM Austraffic Intersection Approach SCATS count % Difference Count West 1733 1694 2.25% South 648 761 -17.44% Orrong Road / Oats Street East 1797 1762 1.95% North 448 433 3.35% West 1880 1853 1.44% Orrong Road / Wright Street East 1850 1784 3.57% North 397 343 13.60% West 1969 1901 3.45% Orrong Road / Archer Street South 520 461 11.35% East 1888 1891 -0.16% West 2249 1861 17.25% South 16 15 6.25% Orrong Road / Francisco Street East 2122 2058 3.02% North 445 - N/A Total 17517 16817 4.00% OFFPEAK 9:00 - 10:00 PM Austraffic Intersection Approach SCATS count % Difference Count West 695 682 1.87% South 169 238 -40.83% Orrong Road / Oats Street East 497 499 -0.40% North 79 76 3.80% West 775 773 0.26% Orrong Road / Wright Street East 515 503 2.33% North 106 96 9.43% West 828 817 1.33% Orrong Road / Archer Street South 188 180 4.26% East 541 524 3.14% West 971 828 14.73% South 7 4 42.86% Orrong Road / Francisco Street East 687 670 2.47% North 139 - N/A Total 6058 5890 2.77%

Aurecon Australia Pty Ltd ABN 54 005 139 873 Level 1 Septimus Roe Square 256 Adelaide Terrace Perth WA 6000 Australia

T +61 8 6145 9300 F +61 8 6145 5020 E [email protected] W aurecongroup.com

Aurecon offices are located in: Angola, Australia, Botswana, China, Ethiopia, Hong Kong, Indonesia, Lesotho, Libya, Malawi, Mozambique, Namibia, New Zealand, Nigeria, Philippines, Singapore, South Africa, Swaziland, Tanzania, Thailand, Uganda, United Arab Emirates, Vietnam.